结合深度图与美学评估的古建筑场景路径规划方法研究

Research on Path Planning Method of Ancient Architectural Scene Based on Depth Map and Aesthetic Evaluation

随着虚拟现实技术的发展,在宣传、介绍旅游景点时越来越广泛的使用了虚拟景点。而漫游功能作为虚拟景点中的重要的功能之一,为了使浏览者了解真实的景点环境,需要根据浏览者的需求及真实的环境进行路径规划。为在保证路径上没有障碍物的前提下找出两点之间足够短的路径且路径上能够浏览到足够赏心悦目的风景。首先主要通过对影像美学规则进行分析,得到影响美学质量的因素。然后对蚁群算法进行改进,加入影像质量约束,同时避免死锁问题,对一个未知场景进行路径规划,得到一条符合影像美学的路径。通过实验表明,该方法得到的路径上平均风景质量相较于蚁群算法更高,更适合古建筑场景。

With the development of virtual reality technology, virtual scenic spots are more and more widely used in promotion and introduction of tourist attractions. Roaming function is one of the important functions in virtual scenic spots. In order to make visitors understand the real scenic environment, it is necessary to plan the path according to the needs of visitors and the real environment, find a short enough path between two points without obstacles on the premise that visitors can browse enough pleasing scenery on the path. Firstly, through the analysis of image aesthetic rules, we obtain the factors that affect the aesthetic quality. Then we improve the ant colony algorithm, add image quality constraints, and avoid deadlock problem. We plan a path for an unknown scene, and get a path that conforms to image aesthetics. Experimental results show that the average landscape quality obtained by this method is higher than that of ant colony algorithm, and this method is more suitable for ancient architectural scenes.